This study investigates the practical effects of room acoustics on the human ability to localize the sources of sound. Room experiments are supplemented by headphone experiments because the goal of the work is to gain a first principles understanding of localization processes. The sound that comes to a listener directly from a source is binaurally coherent, and listeners can use interaural differences in time and level to localize a steady-state sound. Room reflections, however, lead to binaural incoherence. A science of binaural room acoustics is proposed to address the following questions: How much binaural coherence is needed for localization compared to what is available to listeners in rooms? How does binaural incoherence differentially affect the use of interaural time or interaural level information? Does the binaural incoherence model of rooms adequately capture the character of real rooms? What are the perceptual consequences of chaotic deviations from the average incoherence? Listeners can localize transient sounds in a room because of the precedence effect, which suppresses early reflections. Projects of this study ask: What is the role of late reflections and reverberation in suppressing early reflections as they occur in a room? What are the separate roles of interaural time and level differences in the breakdown of the precedence effect that is caused by rapid changes in stimuli? Is it possible for an onset transient to achieve a long term suppression of implausible steady-state localization cues in headphone listening, similar to the Franssen effect that exhibits long term suppression in a room? In addition to the study of room effects, coherence and precedence, this application proposes new technology to synthesize the image of sounds localized in space. It is a variant on current techniques that permit interaural parameters to be varied with unprecedented flexibility and reliability. The technique will be used to supplement the real-room and headphone experiments on binaural coherence and precedence.